ARTICLE

Received 20 Jun 2013 | Accepted 22 Aug 2013 | Published 19 Sep 2013 DOI: 10.1038/ncomms3495 OPEN Structure of a helicase–helicase loader complex reveals insights into the mechanism of bacterial primosome assembly

Bin Liu1,2, William K. Eliason1,2 & Thomas A. Steitz1,2,3

During the assembly of the bacterial loader-dependent primosome, helicase loader proteins bind to the hexameric helicase ring, deliver it onto the oriC DNA and then dissociate from the complex. Here, to provide a better understanding of this key process, we report the crystal structure of the B570-kDa prepriming complex between the Bacillus subtilis loader protein and the Bacillus stearothermophilus helicase, as well as the helicase-binding domain of primase with a molar ratio of 6:6:3 at 7.5 Å resolution. The overall architecture of the complex exhibits a three-layered ring conformation. Moreover, the structure combined with the proposed model suggests that the shift from the ‘open-ring’ to the ‘open-spiral’ and then the ‘closed- spiral’ state of the helicase ring due to the binding of single-stranded DNA may be the cause of the loader release.

1 Department of Molecular Biophysics and Biochemistry, Yale University, New Haven, Connecticut 06520, USA. 2 Howard Hughes Medical Institute, New Haven, Connecticut 06510, USA. 3 Department of Chemistry, Yale University, New Haven, Connecticut 06520, USA. Correspondence and requests for materials should be addressed to T.A.S. (email: [email protected]).

NATURE COMMUNICATIONS | 4:2495 | DOI: 10.1038/ncomms3495 | www.nature.com/naturecommunications 1 & 2013 Macmillan Publishers Limited. All rights reserved. ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3495

he replication of the bacterial chromosome is initiated at exhibit different quaternary structures4,5. The more recent cryo- oriC where the initiator protein DnaA binds to start the EM structure shows a spiral arrangement of the two interacting Tassembly of the enzymatic replisome machine1. The early hexamers with a split ring, consistent with the possibility of a stages of this process involve the assembly of the primosome and ring-breaking role of the loader5. the formation of a functional primosome2,3. Subsequent to the The translocation of the helicase ring in a 50–30 direction on the remodelling of the replication origin induced by DnaA, the oriC transports primase to the initiation sites and enables the assembly of the bacterial loader-dependent primosome occurs in synthesis of RNA primers. Previous studies using gel filtration, discrete steps and involves at least four different proteins (DnaA, fluorescence, cross-linking and atomic force microscopy indicated helicase, helicase loader and primase) that act in a coordinated that primase DnaG binds to helicase to form a functional and sequential manner. primosome with variable stoichiometries that vary from one to In the Escherichia coli system, the helicase loader protein, three primase molecules to one hexameric helicase ring19,31,32. DnaC, complexed with ATP, binds to hexameric helicase DnaB However, our recent structure of BstDnaB in complex with GDP and forms a DnaB6–DnaC6 complex, which has been confirmed and ssDNA substrate suggests that only one primase is included by cryo-electron microscope (cryo-EM) studies4,5. The loader in the complex when it functions in making RNA primers24. The protein delivers the helicase onto the melted DNA single strands DnaG primase consists of three domains: an N-terminal zinc- of the DnaA–oriC nucleoprotein complex at the origin of binding domain, an RNA polymerase domain and a C-terminal replication2,6. In vivo, this delivery is associated with the helicase-binding domain (HBD)33. Functional interactions initiator protein DnaA2,7, whose amino-terminal domain between DnaB helicase and DnaG primase show that DnaG (NTD) is thought to have a role in loading the helicase and stimulates the ATPase and the helicase activities of DnaB19, and helicase loader complex onto the oriC by interacting with helicase DnaB modulates and increases the synthesis of RNA primers34.It DnaB8. After the loader protein dissociates from the helicase ring, has been believed that loader protein would be ejected from the the NTD of the helicase interacts with the carboxy-terminal helicase and loader complex mounted on the oriC before primase domain (CTD) of the primase and forms a functional primosome binds to the NTD of helicase. However, recent studies that show that synthesizes RNA primers9. Primosome assembly in Gram- the existence of a stable complex between helicase, loader protein positive bacteria is different in the details, including that the and primase15, the synthesis of RNA primers in the presence of corresponding helicase is named DnaC in some bacteria such as the loader protein in vitro14,31 and the effects of ribonucleotides Bacillus subtilis, and the loader protein is DnaI; the assembly of and primase on the release of the loader protein from the helicase the helicase and loader protein complex onto the replication and loader complex35 have suggested that the release of loader origin is assisted by a pair of co-loader proteins DnaB and DnaD protein is coupled with the movement of the helicase ring and the in B. subtilis10–13. synthesis of the RNA primer, and not the initial recruitment of The hexameric helicase DnaC in B. subtilis is inactive primase. in the unwinding activity14 and less stable than the E. coli Although biochemical studies have shed light on the DnaB, and can easily form a mixture of different oligomeric mechanisms of primosome assembly, crystal structures of the states12. However, the stable hexameric DnaB helicase in complexes that occur during primosome assembly are quite B. stearothermophilus (BstDnaB) was observed to form a quite limited. Here we report a crystal structure of BstDnaB helicase stable complex with the loader protein DnaI in B. subtilis in complex with BsuDnaI loader protein and the HBD of (BsuDnaI)15,16. The replicative hexameric DnaB helicase is BstDnaG primase at 7.5 Å resolution. This structure shows how composed of two domains that are connected by one flexible the DnaB helicase interacts with the DnaI loader with a linker17,18. The larger CTD exhibits a classical RecA fold, binds stoichiometry of one hexameric helicase ring binding to three the nucleoside triphosphate and translocates the protein on the loader protein dimers and three HBDs of primase DnaG. DNA, whereas the NTD surrounds the single-stranded DNA Moreover, the structure and the proposed model suggest that the (ssDNA) during DNA unwinding and binds the primase19,20. release of helicase loaders may result from the binding of ssDNA DnaB helicase can bind to ssDNA but not to the oriC in the to the helicase ring, which is accompanied by the shift from the absence of loader protein3,21. The crystal structures of various ‘open-ring’ to the ‘open-spiral’, and then the ‘closed-spiral’ state hexameric helicases in complex with ssDNA or single-stranded ofthehelicasering. RNA show a spiral staircase of ssDNA or single-stranded RNA passing through the central channel of the helicase ring22–24. The DnaI helicase loader belongs to the AAA þ family of Results ATPases25,26. The CTD consists of an AAA þ fold that is Structure determination. The complex of BstDnaB helicase, required for nucleotide and ssDNA binding. Its smaller NTD BsuDnaI loader and the HBD of BstDnaG primase was obtained functions as a molecular switch to regulate the binding of DNA16. by first purifying the DnaB–DnaI complex using gel filtration The NTD of this DnaI loader, which contains a novel zinc- chromatography and then adding the HBD of DnaG (Fig. 1). This binding fold, is not homologous to that of E. coli loader protein procedure is the same as previously reported15. Analysis of the DnaC27, and was found to interact with the CTD of the DnaB proteins in these crystals using mass spectrometry and tryptic helicase in the complex of DnaB with DnaI15. The loading of the digestion confirmed the presence of these three different proteins. hexameric helicase ring onto ssDNA can be achieved by the In addition, another crystal form was also obtained under helicase loader or the primase-like domain of the helicase28. Two different conditions and was shown to contain one hexameric different mechanisms of the loader-dependent loading have been DnaB and three HBDs of DnaG by using molecular replacement. proposed, based on biochemical studies: helicase ring-breaking Further, calculation of the Matthews coefficient confirmed the and ring-making12,29,30. In some species, such as E. coli, helicase existence of hexameric DnaB, hexameric DnaI and three HBDs of loader is believed to aid the parting of a subunit interface in the DnaG in an asymmetric unit. stable helicase ring and thereby break the ring. However, in other The structure was determined by molecular replacement. species, such as B. subtilis, the oligomeric helicases can be Initially, subunits or single proteins were used as search models. assembled into a hexamer ring around DNA by the helicase We then built a new model containing as many components as loader, which acts as a ring maker. There are two reported cryo- possible that were positioned using the cryo-EM density map of EM structures of the E. coli helicase–helicase loader complex that the E. coli helicase and helicase loader complex4 (EMD-1017,

2 NATURE COMMUNICATIONS | 4:2495 | DOI: 10.1038/ncomms3495 | www.nature.com/naturecommunications & 2013 Macmillan Publishers Limited. All rights reserved. NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3495 ARTICLE a Peak 2 160 Peak 2 Peak 1 140 Markers (kDa) 250 150 100 120 75 CTD of DnaB 50 37 100 25 20 Peak 1 15

(mAU) 80 10 280 60

OD BstDnaB CTD of DnaI 40 BsuDnal BstDnaB.BsuDnal 20 0 20 40 60 80 100 120 90° 90° Elution volume (ml)

b kers Peak Peak2 Mar1 (kDa) 250 150 60 100 Peak 1 75 50 37 25 40 20

(mAU) 15 10 280 OD 20 BstDnaB.BsuDnal BstDnaB.BsuDnal.HBD Peak 2 0 Figure 2 | Fitting of the CTDs of both BstDnaB and BsuDnaI into the 50 100 150 200 250 300 350 cryo-EM map of the E. coli helicase and helicase loader complex. The Elution volume (ml) CTDs of BstDnaB and BsuDnaI are shown in blue and cyan, respectively. The grey surface of the cryo-EM density map is contoured at 5 s. Figure 1 | The interactions of BstDnaB, BsuDnaI and the HBD of BstDnaG examined by size-exclusion chromatography. (a) Green, red and blue traces represent the migrations of BstDnaB, BsuDnaI and a premixture of BstDnaB and extra BsuDnaI, respectively, on a 16/60 Superdex G200 prep positions of the a-helices using the averaged maps and rigid- grade gel filtration column with a flow rate of 1 ml min À 1. The inserted body refinements. Although the side chains are not observable at SDS–PAGE gel verifies the presence of the BstDnaB–BsuDnaI complex in this resolution, they are included in the refinements, as they peak 1. (b) Blue and red curves show the migrations of a premixture of greatly improve the Rfree value. The NTDs of BsuDnaI were BstDnaB and extra BsuDnaI and a premixture of the BstDnaB–BsuDnaI manually fitted into averaged maps and 2Fo–Fc density maps complex and extra HBD of BstDnaG, respectively, on a 26/60 Superdex generated using the phases that were improved by symmetric G200 prep grade gel filtration column with a flow rate of 1 ml min À 1.The averaging. The final coordinates of the whole model produce an corresponding peak 1 on the gel indicates the formation of a stable Rfactor of 0.379 and an Rfree of 0.392 (Fig. 3 and Supplementary BstDnaB–BsuDnaI–HBD complex. The figure was created using Igor Pro. Table S1). The electron density map clearly shows the a-helices as rods at the current resolution (Fig. 4a–d). The 2Fo–Fc omit map that was calculated after omitting the HBDs of BstDnaG EMDB accession code). The BstDnaB helicase and the CTD of together with the NTDs of BstDnaB in the complex provides BsuDnaI are homologous to the helicase and the CTD of helicase support to the phasing obtained by molecular replacement loader in E. coli, respectively. However, it was believed that there (Supplementary Fig. S1c). The real-space correlation coefficients is a structural difference between the NTD of BsuDnaI and the for fitting five regions of the model into the maps are shown in NTD of E. coli DnaC27. Therefore, a model containing the CTDs Supplementary Table S2. of BstDnaB and BsuDnaI was positioned into the cryo-EM density map (Fig. 2). As the structures of the different crystals containing DnaB and the HBDs of DnaG mentioned above are Overall structure and interactions. The oligomeric protein quite similar to the structure of BstDnaB in complex with the complex whose structure we determined contains two hexameric HBD of BstDnaG (2R6C, PDB accession code), we assumed the proteins, the helicase and the helicase loader, as well as three conformation of the NTDs of DnaB and the HBDs of DnaG HBDs of the primase that are all required in bacterial primosome should be the same as the 2R6C structure18. Consequently, the assembly. The protein stoichiometry in this structure is consistent NTDs of BstDnaB and the HBDs of BstDnaG were added into with that of previously reported structures4,18 or biochemical the model by superimposing the CTDs of the helicase in the studies, including gel filtration analysis12, co-expression and model on those of the BstDnaB and HBD complex structure fusion protein expression of the helicase and the helicase loader36, (2R6C), which generated a new model. The averaged maps surface plasmon resonance experiments using a Biacore16 and (Supplementary Fig. S1a,b) created by refining and averaging fluorescence measurements37. BstDnaB alone without including the HBDs or the loaders from The overall architecture of the 570-kDa complex exhibits a the beginning provide support for the model. three-layered ring structure along the c axis of the crystal (Fig. 3). Initial phases were obtained using the new model that was The NTDs of BstDnaB together with the HBDs of BstDnaG form created using the cryo-EM density as a guide. The phases the first ring that lies on the top of the second ring, which consists calculated from the model were improved through threefold or of the CTDs of BstDnaB. This double-layered ring structure sixfold non-crystallographic symmetry averaging of the different resembles our previous structure of the helicases with HBDs proteins or domains. Because of the modest resolution of the (PDB accession code: 2R6C) with an root mean square deviation X-ray data, coordinates were optimized by the adjustment of the of 2.44 Å between corresponding Ca atoms. The NTDs of

NATURE COMMUNICATIONS | 4:2495 | DOI: 10.1038/ncomms3495 | www.nature.com/naturecommunications 3 & 2013 Macmillan Publishers Limited. All rights reserved. ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3495

c ab b HBD of DnaG a NTD of DnaB

130 Å CTD of DnaB

NTD of DnaI cd CTD of DnaI

126 Å

90° 90°

e CTD NTD

NTD Cleft Cleft CTD 180°

NTD CTD Figure 3 | Views of the overall structure of BstDnaB in complex with Discontiguous side Contiguous side BsuDnaI and HBD of BstDnaG. The surface representations of the Figure 4 | Views of the averaged electron density maps in different parts structure of the complex containing the HBDs of BstDnaG (green), the of the complex and the rings of the helicase loader. The colour NTDs of BstDnaB (yellow), the CTDs of BstDnaB (blue), the NTDs of scheme (a–d) for the proteins is same as in Fig. 3. (a) The blue averaged BsuDnaI (magenta) and the CTDs of BsuDnaI (cyan) are shown along the c map, contoured in at 1.5 s, covers one HBD of BstDnaG (chain I) and one and a axes of crystal lattice. The lengths of the complex along these axes NTD of BstDnaB (chain E). (b) The region close to the RecA fold in chain A are around 130 Å and 126 Å, respectively. of BstDnaB is superimposed on a 1.0 s averaged map in blue. (c) One magenta NTD of BsuDnaI (chain K) is superimposed on the blue map contoured at 1.0 s.(d) The blue averaged map contoured at 1.5 s is BstDnaB pack into a more rigid triangular collar shape that superimposed on one AAA þ fold region of BsuDnaI (chain O). (e) Surface exhibits a threefold symmetry, whereas the CTDs have a relatively representations of the hexameric loader ring as seen in the prepriming loose collar shape with a pseudo sixfold symmetry. The BsuDnaI complex structure viewed along the c axis of crystal lattice. The CTDs of the on the bottom forms the loosest triangular collar of trimer of three dimers are cyan, olive and wheat, respectively. The corresponding dimers with an obvious threefold symmetry as well (Fig. 4e). The NTDs are labelled as light pink, light blue and pale green, respectively. The complex assembly has an outer diameter of 145 Å at the site of the conserved potential DNA-binding sites (Lys160, Lys162, Lys286, Arg289 HBDs of DnaG, 120–126 Å at the other domains and a height of and Arg293) are labelled with dark blue. The left figure is discontiguous 130 Å. The diameter of the central channel is around 50–55 Å side of the ring. The right figure is the contiguous side making contacts with throughout its length, which is wide enough to allow either the CTDs of the helicase ring. The observed cleft in the CTD ring of the double-stranded DNA or ssDNA to pass through the channel. loaders is labelled with red. The Rho helicase complexed with nucleotide and RNA exhibited a ‘closed’ state at the CTDs of the helicase ring rather than the ‘open’ state of the apo-Rho helicase23. The diameter of the inner channel formed by the CTDs of BstDnaB is significantly smaller B1,502 Å2, the packing of the loader ring is probably maintained in the presence of substrates, as is also the case with DnaB by both the interactions between domains and the contacts with complexed with DNA and nucleotide24. the CTDs of the helicase. The architecture of the complex is maintained by the The structure of this complex shows that the CTDs of the interactions among its component proteins. The NTDs of the helicase loader interact directly with the helicase ring, which may hexameric BstDnaB helicase interact with all the HBDs of have important mechanistic implications. This observation is primase BstDnaG using a contact surface area of B2,756 Å2, consistent with the results from previous yeast two-hybrid resulting in a tight interaction. The ring of the CTDs of the experiments, which indicated weak interactions between the helicase was previously reported to be held together primarily CTDs of BsuDnaI and BstDnaB helicase16. In this structure, the through their interactions with the linker regions of the CTDs of BsuDnaI interact with the CTDs of BstDnaB in the helicase18. In this complex, it was also observed that the mainly vicinity of residues 211–212 of one monomer and around hydrophobic interactions between the CTDs and the NTDs of the residues 281–282 of the other monomer in each dimer. These helicase bury a total contact surface area of B2,415 Å2 per residues are not located in the conserved nucleotide-binding sites. hexamer, which contributes the ring packing of the CTDs as well. It is interesting that not all the CTDs of the helicase loader make The CTDs of hexameric BstDnaB helicase interact with both the contact with the helicase. The CTDs of one dimer (chains N and NTDs and the CTDs of the helicase loader BsuDnaI with a total O) exhibit a different orientation from the others and do not contact surface area of B1,745 Å2 per hexamer. As the total directly interact with the CTDs of the helicase. The contact contact surface among the three dimers of the helicase loader is surface area between the NTDs of BsuDnaI and the CTDs of

4 NATURE COMMUNICATIONS | 4:2495 | DOI: 10.1038/ncomms3495 | www.nature.com/naturecommunications & 2013 Macmillan Publishers Limited. All rights reserved. NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3495 ARTICLE

BstDnaB is around 1,197 Å2, but that between the CTDs of a BsuDnaI and BstDnaB is only B548 Å2. Therefore, it appears that the NTDs of BsuDnaI have the major role in recruiting the helicase loader into the prepriming complex.

The loader protein DnaI. Our analysis by gel filtration and multiple-angle light scattering (Supplementary Fig. S2a) has demonstrated that the helicase loader is a mixture of dimers and monomers with a preference for the monomer (B93% monomer À 1 of all samples at a concentration of 10 mg ml ) in solution. This ‘Open-ring’ state ‘Closed-spiral’ state result differs from the previous reports on the overexpression of helicase loaders that concluded it was a monomer in solu- b tion12,25,36; however, it apparently provides support for the observation that the helicase loader dimers bind to the helicase ring. The three dimers of the helicase loader form a ring that exhibits threefold symmetry and contacts the CTDs of the helicase with both their CTDs and NTDs through hydrophobic and polar interactions. The different orientation of the CTDs of one dimer (chain N and O) compared with the other two results in a slightly spiral CTD ring and an obvious cleft that appears in Interface I Interface II the contact region (Fig. 4e). In addition, as shown in Fig. 4e, the previously reported conserved potential DNA-binding residues Figure 5 | The rings of the CTDs of the helicase and views of the (Lys160, Lys162, Lys286, Arg289 and Arg293) on the CTDs of the interfaces between the CTDs of the helicase ring and the helicase loader helicase loader36 are observed in two different orientations. ring. (a) The blue and salmon CTD rings of the helicases are from the The residues on one monomer of the loader dimer face towards structure of apo helicase bound to HBD18 and the structure of BstDnaB in the NTD of the molecule it contacts. These residues might complex with ssDNA24, respectively. (b) Two different interfaces are function together with the nearby NTD in the initial step of displayed. The colour scheme for BsuDnaI is same as that used in Fig. 3. binding DNA. Those on the other monomer face towards the Left: the interface is from the 570-kDa complex structure. The CTDs of inner channel of the loader ring and probably have roles in BstDnaB are labelled with blue. Right: the interface between the salmon further stabilizing and holding DNA. CTDs of BstDnaB and the BsuDnaI obtained by superimposing the new Thecryo-EMstudyofBarcenaet al.4 determined that the ‘closed-spiral’ orientations of the CTDs of BstDnaB into the 570-kDa E. coli helicase loader DnaC interacts asymmetrically with the complex is represented. helicase DnaB. The loader DnaC makes extensive contacts with the hexameric DnaB ring and exhibits a relative twist angle of B50° among DnaC and DnaB dimers. In the present crystal Effects of ATP or/and ssDNA on the prepriming complex.To structure, the two monomers in each dimer of the helicase test this hypothesis for the mechanism of the loader release, we loader make distinct interactions with the helicase. One utilized size-exclusion chromatography to investigate the effects monomer contacts two CTDs of the helicase, whereas the of these substrates on the prepurified DnaB–DnaI complex. We other one interacts with only one CTD of the helicase. These observed that mixing the complex with ATP and ssDNA observations are similar to the conclusions drawn from the cryo- (polyT14) resulted in the release of the loader from the complex EM study. In addition, the two monomers of each dimer of this (Fig. 6). Furthermore, addition of only ATP did not release the loader DnaI do not exhibit twofold symmetry relative to each loader, but mixing the complex with ssDNA alone led to other, as the positions of their NTDs relative to their CTDs the appearance of the loader peak (Supplementary Fig. S2b). The differ. presence of the Mg2 þ did not affect the results of the experi- ments either. These data suggested that BsuDnaI could load BstDnaB onto the ssDNA in the absence of ATP, which is con- The ‘spiral’ states of the helicase ring. Our recent structure of sistent with the previous report21; ATP binding or hydrolysis BstDnaB helicase in complex with GDP and ssDNA demon- might not be required for the loader release at this stage. In strated a distinct non-planar conformation of the helicase ring addition, pre-incubating the mixture at 37 °C for 30 min did not representing the translocation state of the ring24. The CTDs of change the results, which indicated the rapidity of the reaction. BstDnaB helicase in that structure display a new ‘spiral’ Replacing the polyT14 ssDNA with a polyT8 ssDNA gave a conformation, which results in a smaller central channel and similar result. Through analysing relative intensities of the bands tighter packing of the ring (Fig. 5a). Interestingly, when the new on the SDS–polyacrylamide gel electrophoresis (SDS–PAGE) gels, ‘spiral’ orientations of the CTDs of BstDnaB are superposed into it was found that the released BsuDnaI in the peak 2 of Fig. 6 is the 570-kDa complex, obvious clashes between the helicase and around 20% of the all loader proteins. the loader were observed (Fig. 5b). Subsequently, a new cryo-EM structure of the E. coli DnaB in complex with DnaC, but without ssDNA, demonstrated a different spiral conformation of the Discussion helicase and loader. As this spiral conformation has obvious cleft Wickner and Hurwitz38 showed that the E. coli DnaB and DnaC in the two hexamers that could allow ssDNA binding, it is named proteins could form a complex in the presence of ATP. It was the ‘open-spiral’ state, and the former spiral is called the ‘closed- then proposed that DnaC is released from the complex via a spiral’ state. These observations suggest that the binding of sub- mechanism that is dependent on ATP hydrolysis39. Subsequently, strates, including DNA, induces the conformational changes of the E. coli helicase loader DnaC was proven to be a dual ATP/ the helicase ring and may lead to the loader release. ADP switch protein21. It can bind to ATP, but cannot hydrolyse

NATURE COMMUNICATIONS | 4:2495 | DOI: 10.1038/ncomms3495 | www.nature.com/naturecommunications 5 & 2013 Macmillan Publishers Limited. All rights reserved. ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3495

Peak 3 Although the two proposals presented above have been

Markers Peak 1a Peak 1b Peak 2 Peak 3 supported by different biochemical data, our structure of the 200 (kDa) 250 complex and the ‘spiral’ states of the helicase ring in complex 150 100 24 75 with nucleotide and ssDNA and the E. coli DnaB in complex 150 50 with DnaC suggest an alternative perspective on the release of the 37 25 Peak 2 helicase loader. Our results lead us to suggest that the helicase 20 (mAU) 15 ring may undergo the transformations from the ‘open-ring’ state 260 100 10 to the ‘open-spiral’ state and then the ‘closed-spiral’ state during /OD BstDnaB.BsuDnal (280 nm) the process of the complex binding to the oriC, resulting in the 280 BstDnaB.BsuDnal+ATP+ssDNA (280 nm) BstDnaB.BsuDnal (260 nm) loader ring dissociating from the complex. This hypothesis was OD 50 BstDnaB.BsuDnal+ATP+ssDNA (260 nm) tested and was supported by our experiments, which examined Peak 1b Peak 1a the effects of ATP or/and ssDNA on releasing the helicase loader 0 from the prepriming complex. The data clearly demonstrated that 810121416182022in vitro, the binding of the ssDNA, not ATP, to the complex leads Elution volume (ml) to a partial release of the helicase loader. Interestingly, Ioannou et al.16 used surface plasmon resonance experiments to observe Figure 6 | Effects of ATP and ssDNA on the BstDnaB–BsuDnaI complex. that the loader BsuDnaI dissociates from the BstDnaB–BsuDnaI The blue and red dash traces represent the migrations of the complex complex after the helicase is loaded onto ssDNA, which also (40 ml, 7.5 mg ml À 1) on a Superdex 200 10/300 GL column with a flow À 1 supports our proposal for the release of the helicase loader. rate of 0.5 ml min at 280 nm and 260 nm, respectively. The migrations of On the basis of our biochemical data, the structure of the a mixture of the same amount of the complex with 1 mM ATP and three complex of BstDnaB, BsuDnaI and the HBD of BstDnaG, the new times molar ratios of ssDNA (polyT14) without incubation are proposed model, the previously reported data35 and the recent demonstrated with blue and red traces (280 nm and 260 nm, respectively). cryo-EM structures of E. coli DnaB in complex with DnaC4,5,we The SDS–PAGE analysis shows the proteins present in the pooled peaks. can speculate how the loaders are released during the DnaA- Combined with other controls and results (see Methods), it is presumed dependent primosome assembly. that peak 1a is the DnaB–DnaI complex, peak 1b is the complex together The helicase–helicase loader complex exhibits at least two with ATP and ssDNA, peak 2 is the protein BsuDnaI in complex with ATP different quaternary structures in solution. These two different and ssDNA and peak 3 is the ssDNA and ATP. conformations include a spiral helicase in complex with spiral helicase loader5, and the open-ring helicase bound with the open- ring helicase loader4. The former is probably the preferred conformation that allows helicase loading onto ssDNA. Binding ATP in absence of the helicase DnaB or ssDNA. The DnaC-ATP of the complex with this conformation to the oriC would facilitate state promotes assembly of the helicase DnaB onto the oriC and the transformation from the ‘open-ring’ form to the ‘open-spiral’ inhibits the unwinding activity of DnaB, whereas the DnaC-ADP form, and also result in significant changes in the relative state relieves the inhibition of the helicase. Therefore, it was orientations of the CTDs of the helicase ring as discussed above believed that the hydrolysis of the ATP promotes the dissociation (from the ‘open-spiral’ to the ‘closed-spiral’ state). These changes of the helicase loader DnaC from the complex and thereby will alter the contacts between the helicase and the helicase activates the hexameric helicase. loader, and presumably facilitate the disassembly of the helicase However, recent studies have led to an alternative proposal for loader ring, leading to partial dissociation of the loader proteins. when the helicase loader is released from the complex. This partial release of the helicase loader was observed from our Soultanas15 isolated a stable complex containing BstDnaB, biochemical data (Fig. 6 and Supplementary Fig. S2b). The BsuDnaI and BstDnaG by using size-exclusion chromatography primase can then be recruited to form an intermediate ternary and found that this triple protein complex has a poor unwinding complex together with the helicase ring and the helicase loader. activity, but has a good ATPase activity. We purified the 570-kDa This ternary complex was also observed from previous complex for our experiments using similar procedures. Galletto studies15,31,35. The incorporation of NTPs by the new ternary et al.37 utilized fluorescein-labelled E. coli DnaB helicase and a complex creates the RNA primers to be used for the elongation statistical thermodynamic model to observe that the nucleotide- stage of the DNA replication. On the basis of the previous binding sites on the DnaC loaders are not involved in the report35, we presume that with the translocation of the complex stabilization of the complex; and hydrolysis of NTP bound to along the ssDNA and RNA primer synthesis, the remaining DnaB or DnaC is not required for the release of DnaC from the loader proteins in the prepriming complex would be released, DnaB–DnaC complex. Moreover, ssDNA does not affect the which would result in the formation of the functional primosome. binding of DnaC to DnaB. Furthermore, it was observed that in Although the structure of this prepriming complex was derived the presence of the loader, the helicase–primase complex is from a low-resolution map, information that is available from the capable of synthesizing short RNA primers14,31, which suggested structure, including interactions, stoichiometry and binding the possibility that the complex of helicase, helicase loader and models, is an important supplement to elucidate the mechanism primase synthesizes RNA primers. More recently, Makowska- of bacterial primosome assembly, which also has broad implica- Grzyska and Kaguni35 found that the hydrolysis of ATP by E. coli tions for the prokaryotic and even eukaryotic DNA replication. DnaB or DnaC is insufficient for the release of DnaC from a prepriming complex, which was obtained through first incubating Methods prepriming proteins and DNA, and then purifying the complex Protein expression and purification. B. stearothermophilus DnaB and HBD of by gel filtration chromatography. However, addition of both primase DnaG were expressed and purified as described previously18. The helicase primase and ribonucleotides results in the dissociation of DnaC loader protein DnaI fragment was cloned from B. subtilis 168 genomic DNA from the prepriming complex. Therefore, it was suggested that (American Type Culture Collection) using PCR with primers (Integrated DNA Technologies): forward 50-gcgaccatggaaccaatcggccgttcc-30 and reverse the release of DnaC is coupled with the translocation of DnaB 50-gcgactcgagtggatgtcggcggttttctc-30, and inserted into the NcoI–XhoI site of and RNA primer synthesis, rather than resulting from the plasmid pET21d( þ ) (Novagen) and expressed in B834(DE3) (Novagen). The cells hydrolysis of ATP and binding of primase to the complex. growing in Luria broth media with 100 mgmlÀ 1 Ampicillin (Sigma) were induced

6 NATURE COMMUNICATIONS | 4:2495 | DOI: 10.1038/ncomms3495 | www.nature.com/naturecommunications & 2013 Macmillan Publishers Limited. All rights reserved. NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3495 ARTICLE

with 0.1 mM Isopropyl-b-D-thiogalactoside (Sigma) when the OD600 reached 0.6 at helicase ring and the CTDs of the loader DnaI ring, respectively, using 37 °C, and then continued to grow overnight at 20 °C. The collected cell paste was SUPERPOSE46. CCP4 programmes NCSMASK47 and DM48 were then utilized to quickly frozen in liquid nitrogen, resuspended in buffer A (50 mM TRIS pH 8.0, generate corresponding masks and averaged maps. These maps were used to 0.3 M sodium chloride, supplemented with EDTA-free protease inhibitor cocktail manually adjust the positions of the a-helices of the model using COOT. Local tablets (Roche)) and lysed using a microfluidizer. The supernatant of the lysed cell real-space refinement in COOT was also performed in the process. The structure paste solution was filtered with a 0.22-mM filter unit (Millipore) and loaded onto a was refined initially using cycles of rigid-body refinement and followed by few 10-ml Talon superflow metal affinity resin (Clontech) mounted in a 5.0 Â 20 cm2 cycles of restrained refinement using REFMAC 5.0 (ref. 49). The sequence of the Econo column (Bio-rad). After being washed with buffer A using a volume that was CTD of DnaI in the model was changed to that of BsuDnaI using COOT during 20 times the volume of the resin, the resin was eluted with buffer B (50 mM TRIS the process of refinements, based on the analysis of sequence alignment and pH 8.0, 0.3 M sodium chloride, 0.25 M imidazole). The eluate was then loaded onto secondary structure prediction. At this stage, six NTDs of the helicase loader a 16/60 or 26/60 Superdex G200 prep grade gel filtration column (GE Healthcare) proteins were manually built by positioning the domain structure determined equilibrated in buffer C (20 mM TRIS pH 8.0, 50 mM sodium chloride). The previously by NMR (2K7R, PDB accession code) into 2Fo–Fc maps and averaged pooled fractions were concentrated to B20–30 mg ml À 1 for the further complex maps using COOT. The density maps in the regions of DnaI and the CTDs of formation. The final purity of the BsuDnaI sample was over 95% as judged by DnaB appear to be poorer compared with those in the other parts of the complex, SDS–PAGE analysis and Coomassie blue staining. which might be due to the higher content of b-sheets. The data set was collected, processed and refined at 6 Å resolution. However, the outer shell oI/sI4 value is 2.0 at 7.5 Å. All figures were created using PyMOL50. The real-space correlation Isolation of the prepriming complex. First, the B. stearothermophilus helicase coefficients for the fit of each of five domains to both 2Fo–Fc and omit maps were DnaB was mixed with an excess of B. subtilis loader DnaI and directly loaded onto calculated using OVERLAPMAP51. a 16/60 or 26/60 Superdex G200 prep grade gel filtration column (GE Healthcare) with buffer C (Fig. 1a). The fractions containing the DnaB–DnaI complex were À 1 pooled and concentrated to around 10–20 mg ml . Extra B. stearothermophilus Contact analysis and contact area calculation. The intermolecular and inter- HBD of primase DnaG was then added into this complex and loaded onto a 26/60 subunit contacts were analysed using the CCP4 programme CONTACT47 with a Superdex G200 prep grade gel filtration column (GE Healthcare) in buffer C maximum distance of 5 Å. The calculations of the contact areas are performed (Fig. 1b). The fractions containing the complex of DnaB, DnaI and HBD of DnaG using AREAIMOL52 with a probe sphere radius of 1.4 Å. were pooled, concentrated to B10 mg ml À 1 and flash frozen in liquid nitrogen. The final protein complex was over 99% pure as determined by SDS–PAGE ana- lysis and Coomassie blue staining. References 1. Erzberger, J. P., Pirruccello, M. M. & Berger, J. M. The structure of bacterial DnaA: implications for general mechanisms underlying DNA replication Effects of ATP or/and ssDNA on the prepriming complex. The pure DnaB– initiation. Embo J. 21, 4763–4773 (2002). DnaI complex (40–50 ml, 7.5–12 mg ml À 1) was mixed with 1 mM ATP (Sigma) or/and three times molar ratios of ssDNA (polyT14 or polyT8, Integrated DNA 2. Mott, M. L. & Berger, J. M. DNA replication initiation: mechanisms and Technologies), and loaded onto a Superdex 200 10/300 GL column (GE Health- regulation in bacteria. Nat. Rev. Microbiol. 5, 343–354 (2007). care) with or without pre-incubation at a flow rate of 0.5 or 1 ml min À 1 in buffer 3. Fang, L., Davey, M. J. & O’Donnell, M. Replisome assembly at oriC, the C. The ATP and ssDNA were predissolved in 10 mM Tris, pH 8.0. The DnaB–DnaI replication origin of E. coli, reveals an explanation for initiation sites outside an complex, ATP or ssDNA was also loaded individually onto the column as the origin. Mol. Cell 4, 541–553 (1999). negative controls. The reactions were performed in the presence or absence of 4. Barcena, M. et al. The DnaB.DnaC complex: a structure based on dimers assembled around an occluded channel. Embo J. 20, 1462–1468 (2001). 1 mM MgCl2 as well. The pooled peaks were examined using SDS–PAGE analysis and SimplyBlue SafeStain (Invitrogen). Two wavelengths, 280 nm for measuring 5. Arias-Palomo, E., O’Shea, V. L., Hood, I. V. & Berger, J. M. The bacterial DnaC protein and 260 nm for monitoring DNA, were utilized in the experiments. The helicase loader is a DnaB ring breaker. Cell 153, 438–448 (2013). relative intensities of the bands on the SDS–PAGE gels were investigated using 6. Bramhill, D. & Kornberg, A. Duplex opening by dnaA protein at novel KODAK 1D scientific imaging systems. sequences in initiation of replication at the origin of the E. coli chromosome. Cell 52, 743–755 (1988). 7. Carr, K. M. & Kaguni, J. M. Stoichiometry of DnaA and DnaB protein in Crystallization and data processing. Crystals of the prepriming complex were initiation at the Escherichia coli chromosomal origin. J. Biol. Chem. 276, grown using the condition G7 of the PEGs suite (Qiagen) and the sitting drop 44919–44925 (2001). vapour diffusion method. Subsequently, the hanging drop vapour diffusion method 8. Marszalek, J. & Kaguni, J. M. DnaA protein directs the binding of DnaB protein and microseeding techniques were utilized to optimize the crystals. The optimized in initiation of DNA replication in Escherichia coli. J. Biol. Chem. 269, condition used was 11–13% (w/v) polyethylene glycol 3,350, 0.2 M lithium sulphate 4883–4890 (1994). and 50 mM TRIS, pH 8.0–8.3. The complex in the crystal was confirmed by trypsin 9. Tougu, K. & Marians, K. J. The extreme C terminus of primase is required for digestion and protein identification using mass spectrometry (WM Keck Facility at interaction with DnaB at the replication fork. J. Biol. Chem. 271, 21391–21397 the Yale University). In brief, five to six crystals were washed three times in buffer (1996). D (20% (w/v) polyethylene glycol 3,350, 0.2 M lithium sulphate and 50 mM TRIS, 10. Bruand, C., Farache, M., McGovern, S., Ehrlich, S. D. & Polard, P. DnaB, DnaD pH 8.2), dissolved in double-distilled water and then sent for protein identification. and DnaI proteins are components of the Bacillus subtilis replication restart Before flash freezing in liquid nitrogen, crystals were transferred stepwise into a cryo-solution of 30% (w/v) polyethylene glycol 3,350, 0.2 M lithium sulphate and primosome. Mol. Microbiol. 42, 245–255 (2001). 50 mM TRIS, pH 8.2. Data were collected at beamline 24-ID at the Advance 11. Bruand, C. et al. Functional interplay between the Bacillus subtilis DnaD Photon Source and at beamline X-29/X-25 at the National Synchrotron Light and DnaB proteins essential for initiation and re-initiation of DNA replication. Source of Brookhaven National Laboratory. All data were integrated and scaled Mol. Microbiol. 55, 1138–1150 (2005). using the HKL2000 suite of programmes40. 12. Velten, M. et al. A two-protein strategy for the functional loading of a cellular replicative DNA helicase. Mol. Cell 11, 1009–1020 (2003). 13. Briggs, G. S., Smits, W. K. & Soultanas, P. Chromosomal replication initiation Molecular docking and model building. The C-terminal ring of the hexameric machinery of low-G þ C-content Firmicutes. J. Bacteriol. 194, 5162–5170 helicase18 (2R6C, PDB accession code) and the CTD ‘dimer’ of the loader DnaI36 (2012). (2W58, PDB accession code) were used separately as search models to obtain an 14. Rannou, O. et al. Functional interplay of DnaE polymerase, DnaG primase and approximate fit of the complex into the cryo-EM density map of E. coli DnaB6– DnaC helicase within a ternary complex, and primase to polymerase hand-off 4 DnaC6 complex (EMD-1017, EMDB accession code) using MolRep. The fit of the during lagging strand DNA replication in Bacillus subtilis. Nucleic Acids Res. 41, 41 coordinates from the pdb files into the density map by MolRep was then 5303–5320 (2013). 42 optimized using the Chimera package to generate a model containing the CTDs 15. Soultanas, P. A functional interaction between the putative primosomal protein of both hexameric helicase DnaB and hexameric loader DnaI. The N-terminal ring DnaI and the main replicative DNA helicase DnaB in Bacillus. Nucleic Acids of hexameric helicase and the HBD of primase DnaG were added onto the CTDs of Res. 30, 966–974 (2002). BstDnaB that had been fitted into the cryo-EM map using the coordinates of the 43 16. Ioannou, C., Schaeffer, P. M., Dixon, N. E. & Soultanas, P. Helicase binding to 2R6C structure and the programme COOT . DnaI exposes a cryptic DNA-binding site during helicase loading in Bacillus subtilis. Nucleic Acids Res. 34, 5247–5258 (2006). Structure determination and refinement. The structure was determined by 17. Bailey, S., Eliason, W. K. & Steitz, T. A. The crystal structure of the Thermus molecular replacement of the model fitted to the cryo-EM map using PHASER44. aquaticus DnaB helicase monomer. Nucleic Acids Res. 35, 4728–4736 (2007). Before being used in PHASER, the B-factors of the model coordinates that had 18. Bailey, S., Eliason, W. K. & Steitz, T. A. Structure of hexameric DnaB helicase been generated by docking them into the cryo-EM density were set to 20. and its complex with a domain of DnaG primase. Science 318, 459–463 (2007). POINTLESS45 was utilized to check the potential Laue groups. The coordinates 19. Bird, L. E., Pan, H., Soultanas, P. & Wigley, D. B. Mapping protein-protein obtained from PHASER were used to calculate four separate operators, which are interactions within a stable complex of DNA primase and DnaB helicase from responsible for the HBDs of DnaG, the NTDs of the helicase ring, the CTDs of the Bacillus stearothermophilus. Biochemistry 39, 171–182 (2000).

NATURE COMMUNICATIONS | 4:2495 | DOI: 10.1038/ncomms3495 | www.nature.com/naturecommunications 7 & 2013 Macmillan Publishers Limited. All rights reserved. ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3495

20. Mesa, P., Alonso, J. C. & Ayora, S. Bacillus subtilis bacteriophage SPP1 42. Pettersen, E. F. et al. UCSF Chimera--a visualization system for exploratory G40P helicase lacking the N-terminal domain unwinds DNA bidirectionally. research and analysis. J. Comput. Chem. 25, 1605–1612 (2004). J. Mol. Biol. 357, 1077–1088 (2006). 43. Emsley, P. & Cowtan, K. Coot: model-building tools for molecular 21. Davey, M. J., Fang, L., McInerney, P., Georgescu, R. E. & O’Donnell, M. graphics. Acta Crystallogr. D. Biol. Crystallogr. 60, 2126–2132 (2004). The DnaC helicase loader is a dual ATP/ADP switch protein. Embo J. 21, 44. McCoy, A. J. et al. Phaser crystallographic software. J. Appl. Crystallogr. 40, 3148–3159 (2002). 658–674 (2007). 22. Enemark, E. J. & Joshua-Tor, L. Mechanism of DNA translocation in a 45. Evans, P. Scaling and assessment of data quality. Acta Crystallogr. D. Biol. replicative hexameric helicase. Nature 442, 270–275 (2006). Crystallogr. 62, 72–82 (2006). 23. Thomsen, N. D. & Berger, J. M. Running in reverse: the structural basis for 46. Krissinel, E. & Henrick, K. Secondary-structure matching (SSM), a new tool for translocation polarity in hexameric helicases. Cell 139, 523–534 (2009). fast protein structure alignment in three dimensions. Acta Crystallogr. D. Biol. 24. Itsathitphaisarn, O., Wing, R. A., Eliason, W. K., Wang, J. & Steitz, T. A. Crystallogr. 60, 2256–2268 (2004). The hexameric helicase DnaB adopts a nonplanar conformation during 47. Collaborative Computational Project, Number 4. The CCP4 suite: programs for translocation. Cell 151, 267–277 (2012). protein crystallography. Acta Crystallogr. D. Biol. Crystallogr. 50, 760–763 25. Mott, M. L., Erzberger, J. P., Coons, M. M. & Berger, J. M. Structural synergy (1994). and molecular crosstalk between bacterial helicase loaders and replication 48. Cowtan, K. Error estimation and bias correction in phase-improvement initiators. Cell 135, 623–634 (2008). calculations. Acta Crystallogr. D. Biol. Crystallogr. 55, 1555–1567 26.Davey,M.J.,Jeruzalmi,D.,Kuriyan,J.&O’Donnell,M.Motorsandswitches: (1999). AAA þ machines within the replisome. Nat. Rev. Mol. Cell Biol. 3, 826–835 (2002). 49. Murshudov, G. N., Vagin, A. A. & Dodson, E. J. Refinement of macromolecular 27. Loscha, K. V. et al. A novel zinc-binding fold in the helicase interaction domain structures by the maximum-likelihood method. Acta Crystallogr. D. Biol. of the Bacillus subtilis DnaI helicase loader. Nucleic Acids Res. 37, 2395–2404 Crystallogr. 53, 240–255 (1997). (2009). 50. DeLano, W. L. The PyMOL Molecular Graphics System (Delano Scientific, Palo 28. Soultanas, P. Loading mechanisms of ring helicases at replication origins. Alto, CA, USA, 2002). Mol. Microbiol. 84, 6–16 (2012). 51. Branden, C. & Jones, A. Between objectivity and subjectivity. Nature 343, 29. Ahnert, P., Picha, K. M. & Patel, S. S. A ring-opening mechanism for DNA 687–689 (1990). binding in the central channel of the T7 helicase-primase protein. Embo J. 19, 52. Lee, B. & Richards, F. M. The interpretation of protein structures: estimation of 3418–3427 (2000). static accessibility. J. Mol. Biol. 55, 379–400 (1971). 30. Davey, M. J. & O’Donnell, M. Replicative helicase loaders: ring breakers and ring makers. Curr. Biol. 13, R594–R596 (2003). 31. Mitkova, A. V., Khopde, S. M. & Biswas, S. B. Mechanism and stoichiometry of Acknowledgements interaction of DnaG primase with DnaB helicase of Escherichia coli in RNA We thank the staffs at the Advanced Photon Source beamline 24-ID and at the National primer synthesis. J. Biol. Chem. 278, 52253–52261 (2003). Synchrotron Light Source beamline X-29/X-25 for help during data collection. This 32. Thirlway, J. et al. DnaG interacts with a linker region that joins the N- and research was supported by NIH grant GM57510 to T.A.S. T.A.S. is an investigator of the C-domains of DnaB and induces the formation of 3-fold symmetric rings. Howard Hughes Medical Institute. Nucleic Acids Res. 32, 2977–2986 (2004). 33. Frick, D. N. & Richardson, C. C. DNA primases. Annu. Rev. Biochem. 70, 39–80 (2001). Author contributions 34. Thirlway, J. & Soultanas, P. In the Bacillus stearothermophilus DnaB-DnaG B.L. performed experiments and solved the structure. W.K.E. performed the multiple- complex, the activities of the two proteins are modulated by distinct but angle light-scattering analysis. B.L. and T.A.S. wrote the manuscript. overlapping networks of residues. J. Bacteriol. 188, 1534–1539 (2006). 35. Makowska-Grzyska, M. & Kaguni, J. M. Primase directs the release of DnaC from DnaB. Mol. Cell 37, 90–101 (2010). Additional information 36. Tsai, K. L., Lo, Y. H., Sun, Y. J. & Hsiao, C. D. Molecular interplay between the Accession code: The diffraction data and coordinates have been deposited in the Protein replicative helicase DnaC and its loader protein DnaI from Geobacillus Data Bank under accession code 4M4W. kaustophilus. J. Mol. Biol. 393, 1056–1069 (2009). 37. Galletto, R., Jezewska, M. J. & Bujalowski, W. Interactions of the Escherichia Supplementary Information accompanies this paper at http://www.nature.com/ coli DnaB helicase hexamer with the replication factor the DnaC protein. Effect naturecommunications of nucleotide cofactors and the ssDNA on protein-protein interactions and the Competing financial interests: The authors declare no competing financial interests. topology of the complex. J. Mol. Biol. 329, 441–465 (2003). 38. Wickner, S. & Hurwitz, J. Interaction of Escherichia coli dnaB and dnaC(D) Reprints and permission information is available online at http://npg.nature.com/ gene products in vitro. Proc. Natl Acad. Sci. USA 72, 921–925 (1975). reprintsandpermissions/ 39. Wahle, E., Lasken, R. S. & Kornberg, A. The dnaB-dnaC replication protein complex of Escherichia coli. I. Formation and properties. J. Biol. Chem. 264, How to cite this article: Liu, B. et al. Structure of a helicase–helicase loader complex 2463–2468 (1989). reveals insights into the mechanism of bacterial primosome assembly. Nat. Commun. 40. Otwinowski, Z. & Minor, W. Processing of X-ray Diffraction Data Collected in 4:2495 doi: 10.1038/ncomms3495 (2013). Oscillation Mode. in Methods in Enzymology Vol. 276 (eds Carter, C. W. & Sweet, J. R. M.) 307–326 (Academic Press, New York, 1997). This article is licensed under a Creative Commons Attribution 3.0 41. Vagin, A. & Teplyakov, A. MOLREP: an automated program for molecular Unported Licence. To view a copy of this licence visit http:// replacement. J. Appl. Crystallogr. 30, 1022–1025 (1997). creativecommons.org/licenses/by/3.0/.

8 NATURE COMMUNICATIONS | 4:2495 | DOI: 10.1038/ncomms3495 | www.nature.com/naturecommunications & 2013 Macmillan Publishers Limited. All rights reserved.